Generalized Biomedicine Discovery

Luyao Tang, Yingkai Yang, Hanqi Chen, Jiewei Zheng, Chaoqi Chen, Cheng Chen

European Conference on Computer Vision (ECCV), 2026.

Extends category discovery to biomedical data, jointly organizing known and novel concepts in realistic open-world settings.

[Paper] [Code]

Temporal State Transport in Video Generation: Diagnosing and Correcting Spectral Imbalance

Luyao Tang, Bingjun Luo, DONG Yi, Jialin Guo, Haoning Xi, Cheng Chen, Yizhou Yu, Chaoqi Chen

ICML 2026 Workshop — From Frames to Stories (F2S). Best Paper Award · Oral Presentation.

Diagnoses temporal spectral imbalance in video generation and transports internal states to improve long-horizon temporal consistency.

[Paper] [Code]

REFLEX-Med: Reinforcement with Label-Free Explainability for Unified Medical Reasoning

Luyao Tang, Zheyuan Cai, Qinong Tian, Zi Li, Quande Liu, Kyongtae Tyler Bae, Cheng Chen

IEEE International Workshop on Multimedia Signal Processing (IVMSP), 2026. Best Paper Award · Oral Presentation.

Uses label-free rewards for visual fidelity and cross-modal provenance to make unified medical reasoning more accurate and auditable.

[Paper]

CoGe-GCD: Reframing Generalized Category Discovery with Compositional Generalization

Luyao Tang, Jiewei Zheng, Kunze Huang, Chaoqi Chen, Yue Huang, Cheng Chen

Forty-third International Conference on Machine Learning (ICML), 2026.

Reframes generalized category discovery through compositional generalization, transferring reusable visual components from known to novel categories.

[Paper] [Code]

Bures-Isotropy Alignment: Manifold Learning of Generalized Category Discovery

Luyao Tang*, Kunze Huang*, Chaoqi Chen, Cheng Chen

The Fourteenth International Conference on Learning Representations (ICLR), 2026.

Restores isotropic token geometry with a Bures-inspired objective to improve cluster separation and category-number estimation.

[Paper] [Code]

Dissecting Generalized Category Discovery: Multiplex Consensus under Self-Deconstruction

Luyao Tang*, Kunze Huang*, Chaoqi Chen, Yuxuan Yuan, Chenxin Li, Xiaotong Tu, Xinghao Ding, Yue Huang

IEEE/CVF International Conference on Computer Vision (ICCV), 2025.

Decomposes images into visual primitives and combines dominant and contextual consensus for generalized category discovery.

[Paper] [Code]

ASGS: Single-Domain Generalizable Open-Set Object Detection via Adaptive Subgraph Searching

Yuxuan Yuan*, Luyao Tang*, Yixin Chen, Chaoqi Chen, Yue Huang, Xinghao Ding

IEEE/CVF International Conference on Computer Vision (ICCV), 2025.

Searches adaptive object subgraphs and learns compact class embeddings to detect unknown objects across unseen domains.

[Paper]

Generalized Category Discovery via Token Manifold Capacity Learning

Luyao Tang*, Kunze Huang, Chaoqi Chen, Cheng Chen

arXiv preprint, 2025. Under review.

Maximizes class-token manifold capacity to preserve semantic diversity and prevent dimensional collapse during category discovery.

[Paper] [Code]

OCRT: Boosting Foundation Models in the Open World with Object-Concept-Relation Triad

Luyao Tang*, Yuxuan Yuan*, Chaoqi Chen, Zeyu Zhang, Yue Huang, Kun Zhang

IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2025.

Builds sparse object–concept–relation graphs to improve the open-world robustness of foundation models such as SAM and CLIP.

[Paper] [Code]

Reconstruct and Match: Out-of-Distribution Robustness via Topological Homogeneity

Chaoqi Chen, Luyao Tang, Hui Huang

Advances in Neural Information Processing Systems (NeurIPS), 2024.

Learns object components through reconstruction and reasons over their topology for robust out-of-distribution recognition.

[Paper]

Bootstrap Segmentation Foundation Model under Distribution Shift via Object-Centric Learning

Luyao Tang*, Yuxuan Yuan*, Chaoqi Chen, Kunze Huang, Xinghao Ding, Yue Huang

ECCV Workshop on EVAL-FoMo, 2024.

Injects self-supervised object-centric representations into SAM to improve segmentation under distribution shifts.

[Paper] [Code]

Mixstyle-Entropy: Domain Generalization with Causal Intervention and Perturbation

Luyao Tang*, Yuxuan Yuan*, Xinghao Ding, Chaoqi Chen, Yue Huang

British Machine Vision Conference (BMVC), 2024.

Combines causal intervention during training with causal perturbation at test time for more robust domain generalization.

[Paper] [Code]

CODA: Generalizing to Open and Unseen Domains with Compaction and Disambiguation

Chaoqi Chen*, Luyao Tang*, Yue Huang, Xiaoguang Han, Yizhou Yu

Advances in Neural Information Processing Systems (NeurIPS), 2023.

Compacts known-class representations and disambiguates open classes at test time for generalization to unseen domains.

[Paper]

Activate and Reject: Towards Safe Domain Generalization under Category Shift

Chaoqi Chen*, Luyao Tang*, Leitian Tao, Hong-Yu Zhou, Yue Huang, Xiaoguang Han, Yizhou Yu

IEEE/CVF International Conference on Computer Vision (ICCV), 2023.

Activates an explicit unknown response and adapts predictions online for safer recognition under domain and category shifts.

[Paper]

Mix and Reason: Reasoning over Semantic Topology with Data Mixing for Domain Generalization

Chaoqi Chen, Luyao Tang, Feng Liu, Gangming Zhao, Yue Huang, Yizhou Yu

Advances in Neural Information Processing Systems (NeurIPS), 2022.

Combines category-aware data mixing with relational reasoning to preserve semantic topology across domains.

[Paper]